Home  »  How HIVQUAL Works  »  The HIVQUAL Framework  »  Performance Measurement  »  Sampling

Sampling

The HIVQUAL sampling strategy allows just enough patient records to be reviewed to provide statistically significant results.

Samples are generated according to the following methodology:

  1. A case list of active patients is generated by the clinic
  2. Eligible patients are identified
  3. Sample size is determined by active caseload and may be stratified [see below]
  4. From this list of eligible patients a randomized sample is generated
  5. Focused medical record abstraction of selected records is conducted for core indicators
  6. Data are entered into the software and reports are generated automatically
  7. Data are submitted to the Ministry of Health for generation of aggregate reports
SAMPLING TABLE
Population Size Sample Size for 90%
Confidence Interval		 Sample Size for 95%
Confidence Interval
0-49 35 38
50-79 43 48
80-99 52 61
100-139 61 73
140-179 67 82
180-199 70 86
200-249 75 94
250-299 79 101
300-349 82 106
350-749 94 127
750-4999 105 146
5000 or more 107 150
Sample sizes are calculated to achieve a margin of error of 0.8, i.e. a width of 1.6

Stratification:
In some cases, the sample can be stratified as needeed according to the indicator definitions.  Samples can be stratified by gender to account for gynecological indicators, by age for some pediatric indicators, or other relevant criteria.

For example, HIVQUAL-T pediatric samples are stratified by age, as some indicators apply only to specific age groups (e.g., HIV disclosure is only measured in patients > 10 years of age).